On Thursday 31 May 2018 at 22:14:04, Chris Boot wrote: > Hi list, > > In the old days of Nagios and NRPE, monitoring HA clusters was a simple > enough hack: declare a host that represents the services provided by the > cluster, assign a virtual IP address that floats with the services > (which you probably need to have anyway), then have Nagios prod NRPE on > the virtual address. Hey presto you're monitoring your services on > whichever node happens to be active. > > With Icinga2 and its zones and endpoints, this doesn't work. Endpoints > are very specific hosts with SSL certificates, and they can't just > freely float between hosts. > > Is there some way I can replicate this functionality without resorting > to NRPE?
1. I think you need to be careful to distinguish between *servers*, which you want to monitor quite individually, the *services* running on each server, which you might want to monitor with some dependencies to avoid false alerts in the case of a service not running on a cluster machine in standby mode, and the *cluster service* itself (which you monitor from outside the cluster, and don't care which server is providing the service, so long as it responds). Some clusters have one live machine and one hot standby; some have N servers all running side-by side (distributed / load-balanced rather than master-slave failover) where you want to be notified if fewer than X of the N servers are actually operational; there are many different ways in which a "cluster" might need to be monitored, depending on its architecture. 2. There is a "cluster-zone" Icinga2 check command, which may be what you need (if you can find some sufficiently clear documentation about it; I think this is not easy). 3. The standard documentation for "High-Availability Master with Clients" at https://www.icinga.com/docs/icinga2/latest/doc/06-distributed-monitoring/ may be helpful - consider the clients as the same machines as the masters? 4. If all else fails, it's easy enough to reproduce NRPE functionality using SSH :) https://www.icinga.com/docs/icinga2/latest/doc/07-agent-based-monitoring/ I hope some of the above helps, Antony. -- Numerous psychological studies over the years have demonstrated that the majority of people genuinely believe they are not like the majority of people. Please reply to the list; please *don't* CC me. _______________________________________________ icinga-users mailing list icinga-users@lists.icinga.org https://lists.icinga.org/mailman/listinfo/icinga-users
